Managing the heat generated in electronics is a huge problem, especially with the constant push to reduce the size and pack as many transistors as possible in the same chip. The whole problem is how to manage such high heat fluxes efficiently. Usually electronic technologies, designed by electrical engineers, and cooling systems, designed by mechanical engineers, are done independently and separately. But now EPFL researchers have quietly revolutionized the process by combining these two design steps into one: they've developed an integrated microfluidic cooling technology together with the electronics, that can efficiently manage the large heat fluxes generated by transistors. Their research, which has been published in Nature, will lead to even more compact electronic devices and enable the integration of power converters, with several high-voltage devices, into a single chip.
"This cooling technology will enable us to make electronic devices even more compact and could considerably reduce energy consumption around the world," says Matioli. "We've eliminated the need for large external heat sinks and shown that it's possible to create ultra-compact power converters in a single chip. This will prove useful as society becomes increasingly reliant on electronics." The researchers are now looking at how to manage heat in other devices, such as lasers and communications systems.
[Abstract]: Co-designing electronics with microfluidics
[Source]: EPFL
Journal Reference:
Remco van Erp, Reza Soleimanzadeh, Luca Nela, Georgios Kampitsis, Elison Matioli. Co-designing electronics with microfluidics for more sustainable cooling. Nature, 2020; 585 (7824): 211 DOI: 10.1038/s41586-020-2666-1
(Score: 2, Interesting) by fakefuck39 on Thursday September 10 2020, @11:37PM (11 children)
This is dated 2020, but I'm 100% sure I read about it a couple of years ago.
here's an paper on this from 2006.
https://www.microcooling.com/pdfs/Penn_State_Univ_Article.pdf [microcooling.com]
(Score: 3, Insightful) by takyon on Friday September 11 2020, @01:30AM (2 children)
Was it this one from 2017? Purdue University Shows Off Microchannel Array Chip Cooling Technique [soylentnews.org]
It can take decades for this stuff to become practical. Like how Samsung made a 3D TSV SRAM [anandtech.com] announcement a couple weeks ago but you can find a 2011 paper [ieee.org] about it. A "single-atom transistor" [wikipedia.org] was demoed in 2004, etc.
[SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
(Score: 2) by c0lo on Friday September 11 2020, @03:10AM
These guys beat the "almost 1000W/cm2" achieved by the Purdues.
If you trust the abstract, they got it dialed up to
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford
(Score: 1) by fakefuck39 on Friday September 11 2020, @04:21AM
yep. that's the one. thing is, these aren't related papers. the 2006 paper, the 2017 paper, the 2020 paper - all different research, and they're all announcing it as something new and groundbreaking. which is complete bs. and not bs as in they don't know. bs as in they're lying, because this is all available on the first page of a google search for microchannel cooler or something similar.
(Score: 0) by Anonymous Coward on Friday September 11 2020, @01:55AM (6 children)
"a paper" not "an paper" I'm not a grammar nazi. I'm concerned for the english speaking human race.
(Score: 0) by Anonymous Coward on Friday September 11 2020, @02:43AM (2 children)
Yeah but what about the English speaking human race?
(Score: 0) by Anonymous Coward on Friday September 11 2020, @12:39PM (1 child)
They can go fuck themselves, they gave us this Trump v Clinton and Trump v Biden bullshit.
(Score: 1) by fakefuck39 on Friday September 11 2020, @07:52PM
*whoosh
(Score: 0, Flamebait) by fakefuck39 on Friday September 11 2020, @04:16AM (2 children)
I'm actually concerned for the human race as well. as I sat there on the toilet listening to franceinfo (that's french news, in french), and took a giant dump, I typed with my thumb "an article." I pushed out a hard long turd made out of last night's beef jerkey, and wiped the sweat off my face. I then changed "article" to "paper" with my thumb, and clicked submit.
you see, my concern comes from this. what if more and more people are like you every year? we'll literally die out. you're too dumb to understand writing styles between textbooks, poems, a random website, and a work email. with only one book of rules - the strictest one, you actually go back and proofread your posts to random websites and your text messages. propper grammer must after all be proper, and we must dedicate our time to it and care, because we really have nothing else to do. when you see an error that a 2nd grader wouldn't make, you assume it's becase people actually don't know the rules, instead of just not brothering to check for typos of shit they type on the shitter with their thum. they need to go back and corect they're errors.
then the entire world becomes you - dumb people who are so dumb they think others are dumb. girls don't want anything to doo with losers like that, and then the human race dies out... big consern.
if you really think anyone on this website doesn't know when to use 'a' vs 'an,' you are a complete idiot. and you correct these 2nd grade grammar errors to boost your self esteem. because you can't correct anything that requires any skill or actual knowledge beyond 2nd grade. because you're a moron.
(Score: 0) by Anonymous Coward on Saturday September 12 2020, @12:37AM (1 child)
For the past few years I've seen A LOT of misuse of, 'an,' and, 'a'. I figured out a while ago that apparently British folks will say things like, 'an 'istoric,' event, or, 'an 'our,' ago; which still doesn't sit right with me; but, their British, it can't be helped (j/k). However, I do believe it has extended quite beyond that. Maybe as you say it just has to do with a lot of people shitting and using their smartphone at the same time. In which case. Well. That's shitty. Sorry for shitting all over you so much about that. I've just been holding it in for so long; ya know? Just had to blow it all out at once. Sorry it had to be you; rofl.
(Score: 1) by fakefuck39 on Saturday September 12 2020, @12:49AM
quick question - did you just not learn english, and not very well? an historic, and an hour is correct for both british and all other forms of english. it's only incorrect if you speak southern redneck and wave around a confederate flag. i bet you pronounce the h in who and whip. they made a family guy episode making fun of you mr matlock.
(Score: 2) by c0lo on Friday September 11 2020, @03:07AM
Well, you'll see the problem on page 2
If you aren't sharing, you're fair game to others rediscovering your "proprietary" methods.
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by looorg on Friday September 11 2020, @01:07AM (1 child)
While I like my chips cold to I do have some points of concern. Such as do you need to refill the liquids or will they eventually dry out -- possibly as some form of planned obsolescence? If so then this tech would just be dead on arrival as far as I'm concerned. So will this be some kind of heat-pipe then or chambers inside the chips? I don't think I got that part from the summary. But if they can do it without a heatsink and fans I'm all for it as I value the silence over the high-pitched whine and humming of fans.
(Score: 2) by c0lo on Friday September 11 2020, @03:03AM
Most certainly not. You never do it for heat pipes
If you were to do it, you'd likely need to end using a descale solution on your CPU.
Go read the abstract
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford
(Score: 0) by Anonymous Coward on Friday September 11 2020, @01:30AM
That combo is not that rare.
Even a straight EE needs to learn good bit of ME (in the end, electronic devices are also mechanical/physical devices), and most MEs learn good bit about control system, i.e., electronics.
(Score: 3, Touché) by c0lo on Friday September 11 2020, @02:54AM
From TFAbstract
So, heh, overclock it too high and your CPU will cause a(n overheated) steam explosion!
Can't wait to see the photos.
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford
(Score: 3, Interesting) by VanessaE on Friday September 11 2020, @11:46AM (5 children)
OK, maybe my info is out of date, but isn't the majority of heat generated in any modern IC a byproduct of the underlying CMOS tech itself, in that as one transistor changes states, its complementary circuit takes on the opposite state, so that for a brief moment, the two power rails get shorted together, hence CPUs et al. draw more power and run hotter as their clock speeds increase?
Sure, I know drive voltage changes as clock speed goes up, and the circuit paths and components inside any IC have some degree of resistance, and that every watt used has to eventually end up as heat, but it just seems odd that some new tech has yet to come about that avoids the complementary-shorting thing?
(Score: 1, Interesting) by Anonymous Coward on Friday September 11 2020, @02:00PM
CMOS (as you described) is the simplest and most efficient (parts wise) way of implementing logic gates.
It is lower power than all the preceding digital logic technologies. It is not going anywhere. Any digital logic circuit must consume some power when you change its state.
(Score: 0) by Anonymous Coward on Friday September 11 2020, @06:22PM (1 child)
(Score: 3, Informative) by c0lo on Saturday September 12 2020, @02:36AM
Compute the current caused by switching on/off the gate of a CMOS as a function of frequency. It's quite simple: you have a small capacitor which you charge and discharge over a switch cycle.
Say, Ne electrons is enough to block the gate: you'll have the net charge over one switch cycle at 1.6e-19* Ne/gate/cycle.
Over 1 second that will create a current of 1.6e-19* Ne*switching_freq A.
geforce RTX-3080 has 28 billions of them [nvidia.com] (28e+9) and can work at a switching frequency 2.1GHhz [techspot.com] (2.1e+9).
Assuming that roughly 1/3 of them are charged/discharged over one cycle, that gives a current of 1.96e+19*1.63-19*Ne A = 3.136*Ne A.
If you assume just Ne = 50 electrons are pushed/drained from the gate every switching cycle, you get a current of 156.8A. At a voltage of 3.3V, that gives a consumed power of 517.44 W.
---
One on top of the other, I don't see how one could revolutionize the transistor to do better; go in the range of using just 10 electrons/cycle to keep your state and you're going to deal with quantum mechanics - your computations may or may not be accurate, depending on the whim of the Schrodinger's cat.
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford
(Score: 2) by Rupert Pupnick on Saturday September 12 2020, @01:17AM
Most of the heating in CMOS ICs is due to the currents that flow in all of the parasitic capacitances that charge and discharge during a state change. Having both FETs on at same is avoided due to the risks of blowing up FETs, and since there are millions of FETs on a typical IC, designers need to build in timing margins to the gate drive that can yield functioning product. Also note that in addition to hurting reliability, “shoot through” losses would result in heating that would increase in proportion with the number of transistors on a die. Capacitive losses, on the other hand, get smaller per transistor as the feature size gets smaller which makes it possible to scale up transistor density to what we have today.
(Score: 2) by c0lo on Saturday September 12 2020, @01:54AM
Nope, no SC necessary as an explanation, the flow of the current for functioning under normal conduction is enough.
In order for a transistor to change state, it must shift an amount of charge - that shifting creates a current and the flow of charge will create heat (even more so in case of semi-conductors which, you know, aren't quite as good as the conductors at ... umm.. conducting).
Switching it frequently means you need to transfer more charge from one gate to the other in the same amount of time to effect the state change you need for the computation. If you don't supply enough current, the state of your gates are going to be unpredictable, a thing that may be acceptable for MBA graduates and Trump, but you don't want it happening in your computer.
https://www.youtube.com/@ProfSteveKeen https://soylentnews.org/~MichaelDavidCrawford